Research
Research within the group currently concentrates on enhancing the security of general software/applications, compilers/interpreters, and machine learning systems (AI compilers). This is primarily achieved through the development and application of scalable, cost-effective program analysis (source/binary) and fuzzing techniques.
Security for General Software(Applications)
- Employ program analysis (source/binary) and fuzzing to identify and mitigate security vulnerabilities present in software developed using multiple programming languages.
- PolyCruise, PolyFuzz
- FSE’22, USENIX Security’22, USENIX Security’23
Security for Compilers/Interpreters
- Employ (non)directed fuzzing, machine learning and program analysis (source/binary) to generate realistic applications for mining security vulnerabilities in compilers, interpreters and language runtimes.
- PyRTFuzz
- CCS’23
Security for Machine Learning Systems (AI Compilers)
- Employ fuzzing, program analysis (source/binary) and high-performance computing to identify and mitigate security vulnerabilities in machine learning systems.
